Correlation-Based Parking Availability Estimation

The patent introduces correlation models as intermediaries that transfer parking availability information from monitored parking blocks to un-monitored parking blocks. These models act as mediators that enable indirect observation of parking status in areas without direct sensor coverage, thereby extending system coverage without requiring physical sensors in every location.

The patent creates virtual copies of parking availability data by using correlation models to replicate the parking status information from monitored blocks to un-monitored blocks. Instead of directly measuring parking availability in un-monitored areas, the system generates estimated copies based on patterns observed in monitored areas, enabling coverage extension through data replication.

The patent applies partial action by using only the subset of correlation models that are relevant and accurate for specific parking block pairs. Instead of forcing a single universal model, the system selectively applies multiple specialized models depending on the specific estimation scenario, thereby maintaining high accuracy while extending coverage to various un-monitored areas.

The patent changes the parameters of model selection by dynamically choosing which correlation model to apply based on the specific characteristics of the source and target parking blocks. The system adjusts model parameters such as correlation strength, data recency, and spatial proximity to optimize estimation accuracy for each specific case, thereby maintaining precision across diverse coverage scenarios.

The patent segments the large set of correlation models into smaller, manageable subsets based on spatial relationships, parking block characteristics, and data availability. By dividing the model set into organized groups, the system can efficiently select and apply only the relevant models for each estimation task, reducing computational complexity while maintaining comprehensive coverage and accuracy.

The patent implements dynamic model selection where the system adaptively chooses which correlation model to apply based on current conditions such as data recency, correlation strength, and parking block characteristics. This dynamic approach allows the system to optimize between model complexity and estimation accuracy in real-time, selecting the most appropriate model for each specific situation rather than using a fixed complex set.